Evaluation of the Bifactor Confirmatory Factor Analysis Model of Dimensionality and Value of Sub-Domains of Psychological Scale

The application of methods such as Structural Equation Modeling (SEM) and Item Response Theory (IRT) for the operationalization of psychological constructs necessitates an examination of the assumption of "local independence" within the data. The objective of this research was to assess the dimensionality and additional value of the subdomains of psychological scales through the use of a bifactor confirmatory factor analysis model.

This study employed a review methodology utilizing the bifactor confirmatory factor analysis model, implemented through the BifactorIndicesCalculator statistical package in R software version 4.3.3. The aim was to evaluate the dimensionality and added value of sub-domains of psychological scales. Data were collected from 512 students at the Islamic Azad University, Garmsar branch, during the 2022-2023 academic year. Participants were selected using an availability sampling method and completed the questions from Yousefi's Good Life Scale.

If the total explained common variance index is less than 0.60, the scale is considered multidimensional. Conversely, if it is 0.90 or higher, the scale is deemed sufficiently unidimensional. Additionally, if both the total explained common variance index and the percentage of uncontaminated correlations are at least 0.70, the scale is also regarded as sufficiently unidimensional. Furthermore, if the total explained common variance index is at least 0.60, it is essential for both the percentage of uncontaminated correlations and the hierarchical omega coefficient to be a minimum of 0.70 for the scale to be classified as sufficiently unidimensional.

The results of the study indicate that researchers can utilize the bifactor confirmatory factor analysis model to assess the dimensionality and significance of the subdomains within psychological scales.